Magnetic Structure and Magnetic Properties of ... - ACS Publications

Dec 1, 2006 - Lindgrenite, Cu3(OH)2(MoO4)2, is a hard ferrimagnet with a TC of 13 K, .... Basudev Swain , Duk-Hee Lee , Jae Ryang Park , Chan-Gi Lee ...
0 downloads 0 Views 318KB Size
Inorg. Chem. 2006, 45, 10938−10946

Magnetic Structure and Magnetic Properties of Synthetic Lindgrenite, Cu3(OH)2(MoO4)2 Serge Vilminot,*,† Gilles Andre´,‡ Mireille Richard-Plouet,§ Franc¸ oise Boure´e-Vigneron,‡ and Mohamedally Kurmoo*,# Groupe des Mate´ riaux Inorganiques, IPCMS, CNRS-ULP UMR7504, 23 rue du Loess, BP 43, 67034 Strasbourg Cedex 2, France, Laboratoire Le´ on Brillouin, CEA-CNRS, CEA Saclay, 91191 Gif-sur-YVette Cedex, France, Institut des Mate´ riaux Jean Rouxel, Laboratoire de Chimie des Solides, 2 rue de la Houssinie` re, BP 32229, 44322 Nantes Cedex, France, and Laboratoire de Chimie de Coordination Organique, CNRS-UMR7140, Tectonique Mole´ culaire du Solide, UniVersite´ Louis Pasteur, Institut Le Bel, 4 rue Blaise Pascal, 67000 Strasbourg Cedex 01, France Received June 28, 2006

Synthetic Cu3(OH)2(MoO4)2 consists of Cu3(OH)2 brucite ribbons of edge-sharing copper octahedra connected by MoO4 into a 3D network as in the mineral, lindgrenite, for all temperatures between 1.5 and 300 K. Each ribbon consists of a triangular connection between two different types of copper atom (Cu(1) and 2 Cu(2)) via µ3-OH. The MoO4 acts both as one- and three-atom bridges to connect six Cu atoms belonging to three adjacent ribbons. The magnetic properties are consistent with those of ferrimagnetic chains, and the resulting moment of each chain is parallel below the long-range magnetic ordering at 13 K. The Curie constant is 0.468(1) emu K mol-1 of Cu; the Weiss temperature is −14.2(2) K, and the saturation magnetization at 2 K in 50 kOe is 0.41 N µB mol-1 of Cu. Analyses of the neutron powder diffraction reveal an ordered magnetic state where the moment of Cu(1) is antiparallel to those of the two Cu(2); all of them point along the a axis without any sign of geometrical frustration. Any degeneracy that may be present because of the triangular topology of the Cu atoms (s ) 1/2) appears to be lifted by the distortion from an ideal equilateral geometry of the triangle. The entropy, estimated from the heat capacity measurements, attains 50% of the total of 17.7 J K-1 mol-1, close to that expected for three Cu atoms (3R ln 2), up to the long range ordering temperature, and the remaining is associated with the low dimensionality of the material.

Introduction Synthetic routes for organic, inorganic, and metal-organic materials using the hydrothermal technique are being developed for academic research purposes, as well as for largescale industrial production.1 The ease of these syntheses at modest temperatures, pressures, and also, costs makes the technique very appealing. We and other researchers have * To whom correspondence should be addressed. Telephone: 00 33 3 88 10 71 28. Fax: 00 33 3 88 10 72 47. Email: [email protected] (S.V.). Telephone: 00 33 3 90 24 13 56. Fax: 00 33 3 90 24 13 25. Email: [email protected] (M.K.). † IPCMS, Strasbourg. ‡ Laboratoire Le ´ on Brillouin, Gif-sur-Yvette. § Institut des Mate ´ riaux Jean Rouxel, Nantes. # Universite ´ Louis Pasteur, Strasbourg. (1) (a) Cundy, C. S.; Cox, P. A. Chem. ReV. 2003, 103, 663-701 and references therein. (b) Barrer, R. M. Hydrothermal Chemistry of Zeolites; Academic Press: London, 1982. (c) Rabeneau, A. Angew. Chem., Int. Engl. 1985, 24, 1026.

10938 Inorganic Chemistry, Vol. 45, No. 26, 2006

been involved in reproducing the temperature and pressure conditions usually involved in the elaboration of synthetic minerals.2-4 The preparation of the synthetic compounds appears interesting when studying the magnetic properties since the stoichiometry and purity can be controlled, whereas (2) (a) Vilminot, S.; Richard-Plouet, M.; Andre´, G.; Swierczynski, D.; Boure´e-Vigneron, F.; Marino, E.; Guillot, M. Cryst. Eng. 2002, 5, 177. (b) Vilminot, S.; Richard-Plouet, M.; Andre´, G.; Swierczynski, D.; Guillot, M.; Boure´e-Vigneron, F.; Drillon, M. J. Solid State Chem. 2003, 170, 255. (c) Vilminot, S.; Richard-Plouet, M.; Andre´, G.; Swierczynski, D.; Boure´e-Vigneron, F.; Kurmoo, M. Dalton Trans. 2006, 1455-1462. (d) Ben, Salah, M.; Vilminot, S.; Andre´, G.; Richard-Plouet, M.; Boure´e-Vigneron, F.; Mhiri, T.; Kurmoo, M. Chem.sEur. J. 2004, 10, 2048. (e) Ben Salah, M.; Vilminot, S.; Mhiri, T.; Kurmoo, M. Eur. J. Inorg. Chem. 2004, 2272. (f) Vilminot, S.; Richard-Plouet, M.; Andre´, G.; Swierczynski, D.; Boure´e-Vigneron, F.; Kurmoo, M. Inorg. Chem. 2003, 42, 6859. (g) Ben Salah, M.; Vilminot, S.; Andre´, G.; Richard-Plouet, M.; Mhiri, T.; Takagi, S.; Kurmoo, M. Chem. Mater. 2005, 17, 2612. (h) Ben Salah, M.; Vilminot, S.; Andre´, G.; Richard-Plouet, M.; Boure´e-Vigneron, F.; Mhiri, T.; Kurmoo, M. J. Am. Chem. Soc. 2006, 128, 7972.

10.1021/ic061182m CCC: $33.50

© 2006 American Chemical Society Published on Web 12/01/2006

Magnetic Structure and Properties of Lindgrenite

the minerals may contain different transition metals.2-5 In many cases, the synthetic equivalents of minerals have been obtained, but nature appears to be an excellent, and even better, laboratory since many of the natural minerals have so far not been reproduced. For instance, the anhydrous copper hydroxysulfates, such as Cu3(OH)4SO4, antlerite, and Cu4(OH)6SO4, brochantite, can be easily prepared by hydrothermal synthesis.2 However, the syntheses of the hydrated minerals, such as Cu4(OH)6SO4(H2O)2, langite, posnjakite, and wroewolfeite,6-8 and Cu7(OH)10(SO4)2(H2O)3, schulenbergite, have not been successful at the laboratory level.9 All the hydrated phases and brochantite exhibit a structure build up of brucite-like sheets of distorted octahedral copper. In contrast, antlerite contains triple chains of copper octahedra. In addition to the preparation of known phases, it is possible to replace some of the components or even functionalize the minerals with organic or inorganic moieties with specific properties.10-12 For example, the substitution of sulfate by selenate allows one to obtain the compounds corresponding to antlerite and brochantite,11,12 even if there are no natural equivalents. Among other tetrahedral anions, molybdate is known to form basic salts of copper (i.e., szecnicsite, Cu3(OH)4MoO4, and lindgrenite, Cu3(OH)2(MoO4)2).13,14 Again, the hydrothermal synthesis we performed does not yield the formation of szecnicsite but, instead, produces lindgrenite in a wide spectrum of starting compositions and for different temperatures. Szecnicsite is not isostructural to antlerite, but it contains similar triple chains of edge-shared copper octahedra. Therefore, it would have been of interest to compare the corresponding magnetic properties. On the other hand, lindgrenite also contains chains of edge-shared copper octahedra, with the alternating succession of one then two copper octahedra along the ribbons. The related structure appears interesting from a magnetic point of view, similar to that we experienced for a similar cobalt ribbon in Co3(OH)2(C4O4)2·3H2O.15 Such systems have been of recent interest in the physics community, and the

physicists describe them as linear chain distorted diamonds.16 Moreover, a triclinic modification has been shown17 that differs from the usual monoclinic form in that the ribbons are now parallel to each other rather than being arranged in a herringbone fashion, as for lindgrenite. In their ongoing studies to understand the magnetism of geometrically frustrated systems, Shores et al. have recently reported the synthesis of lindgrenite and two other forms that were elegantly modified by bipyridine and piperazine.3,10 The two ditopic ligands, interestingly, break the inorganic 3D network into layers, consequently transforming the ferromagnetic lindgrenite into antiferromagnets with a critical metamagnetic field of 2 kOe. Although the Weiss constant of lindgrenite is negative and the saturation magnetization is only ∼1/3 of that expected for a ferromagnet, it was described as a ferromagnet. They propose a model-of-spin arrangement, where only the spins within one chain were considered, to explain the observation of metamagnetism in the organic-modified forms.10 Other interest in materials containing molybdenum oxide is the result of their promising applications in the area of catalysis, sorption, electrical conductivity, magnetism, photochemistry, sensors, and energy storage.18-22 However, the majority of these studies are concerned with compounds like FeMoO4 and Fe2(MoO4)3 for the synthesis of formaldehyde or Fe-Mo-O catalysts for the oxidation of toluene to benzaldehyde,23 NiMoO424 or Cu, Mn, and Co molybdates25 for the oxidative dehydrogenation of propane, and Cu3Mo2O9 for catalytic deep oxidation (combustion) of the emitted harmful gases (e.g., CO, NO, and hydrocarbons).26 Interestingly, lindgrenite has been considered for smoke suppression and fire retardation of poly(vinyl chloride).27 Here, we present the hydrothermal synthesis of the hydrogenated and deuterated analogues of lindgrenite, their thermal and optical properties, and magnetic properties as a function of temperature and field. We also report the magnetic structure obtained from neutron powder diffraction and the heat capacity in zero field. All results point to a

(3) (a) Nocera, D. G.; Bartlett, B. M.; Grohol, D.; Papoutsakis, D.; Shores, M. P. Chem.sEur. J. 2004, 10, 3850 and references therein. (b) Wills, A. S. Phys. ReV. B. 2001, 63, 064430. (c) Shores, M. P.; Nytko, E. A.; Bartlett, B. M.; Nocera, D. G. J. Am. Chem. Soc. 2005, 127, 13462. (4) Harrison, A. J. Phys: Condens. Matter 2004, 16, S553. (5) (a) Hawthorne, F. C.; Krivovichev, S. V.; Burns, P. C. ReV. Mineral. 2000, 40, 1. (b) Eby, R. K.; Hawthorne, F. C. Acta Crystallogr. 1993, B49, 28. (6) Galy, J.; Jauf, J.; Pulou, R.; Sempere, R. Bull. Mine´ ral. 1984, 107, 641-648. (7) Mellini, M.; Merlino, S. Z. Kristallogr. 1979, 149, 249-257. (8) Hawthorne, F. C.; Groat, L. A. Am. Mineral. 1985, 70, 1050-1055. (9) Mumme, W. G.; Sarp, H.; Chiappero, P. J. Arch. Sci. GeneVe 1994, 47, 117-124. (10) Shores, M. P.; Bartlett, B. M.; Nocera, D. G. J. Am. Chem. Soc. 2005, 127, 17986-17987. (11) Giester, G.; Monatsh. Chem. 1991, 122, 229-234. (12) Lachenal, G.; Vignalou, J. Thermochim. Acta 1987, 111, 195. (13) (a) Stolz, J.; Armbruster, T. Neues Jahrb. Mineral., Monatsh. 1998, 278-288. (b) Francis, C. A.; Pitman, L. C.; Lange, D. E. Mineral. Rec. 1997, 28, 387-394. (14) (a) Hawthorne, F. C.; Eby, R. K. Neues. Jahrb. Mineral., Monatsh. 1985, 5, 234-240. (b) Calvert, L. D.; Barnes, W. H. Can. Mineral. 1957, 6, 31-51. (c) Barnes, W. H. Am. Mineral. 1949, 34, 163-172. (d) Palache, C. Am. Mineral. 1935, 20, 484-91. (15) Kurmoo, M.; Kumagai, H.; Chapman, K. W.; Kepert, C. J. Chem. Commun. 2005, 3012-3014.

(16) (a) Okamoto, K.; Tonegawa, T.; Kaburagi, M. J. Phys. Condens. Matter 2003, 15, 5979-5995. (b) Kikuchi, H.; Fujii, Y.; Chiba, M.; Mitsudo, S.; Idehara, T.; Tonegawa, T.; Okamoto, K.; Sakai, T.; Kuwai, T.; Ohta, H. Phys. ReV. Lett. 2005, 94, 227201. (17) Xu, Y.; Lu, J.; Goh, N. K. J. Mater. Chem. 1999, 9, 1599-1602. (18) Chippindale, A. M.; Cheetham, A. K. In Studies in Inorganic Chemistry; Elsevier: Amsterdam, 1994; Vol. 19, pp 1146-184 and references therein. (19) Johnson, L. F.; Boyd, G. D.; Nassau, K.; Sode, R. R. Phys. ReV. 1962, 126, 1406. (20) Bruce, D. W.; Hare, D. O. Inorganic Materials; Wiley: Chichester, U.K.; 1992. (21) Bein, T., Ed. Supramolecular Architecture: Synthetic Control in Thin Films and Solids; ACS Symposium Series 499; American Chemical Society: Washington, DC, 1992. (22) Cheetham, A. K. Science 1994, 264, 794 and references therein. (23) Zhang, H.; Shen, J.; Ge, X. J. Solid State Chem. 1995, 117, 127135. (24) Mazzochia, C.; Aboumrad, C.; Diagne, C.; Tempesti, E.; Herrman, J. M.; Thomas, G. Catal. Lett. 1991, 10, 181. (25) Palacio, L. A.; Echavarria, A.; Sierra, L.; Lombardo, E. A. Catal. Today 2005, 107-108, 338-345. (26) Hasan, M. A.; Zaki, M. I.; Kumari, K.; Pasuputely, L. Thermochim. Acta 1998, 320, 23-32. (27) Starnes, W. H., Jr; Pike, R. D.; Cole, J. R.; Doyal, A. S.; Kimlin, E. J.; Lee, J. T.; Murray, P. J.; Quinlan, R. A.; Zhang, J. Polym. Degrad. Stab. 2003, 82, 15-24.

Inorganic Chemistry, Vol. 45, No. 26, 2006

10939

Vilminot et al. ferrimagnetic ground state below 13 K without any sign of geometrical frustration. Experimental Section Preparation of Cu3(OH)2(MoO4)2. Synthetic lindgrenite, Cu3(OH)2(MoO4)2 (1), was obtained using an approach similar to that of Clearfield et al.,28 via the hydrothermal reaction of copper nitrate hemipentahydrate, Cu(NO3)2‚2.5H2O (Aldrich, 98%), sodium molybdate dihydrate, Na2MoO4·2H2O (Aldrich, 99.5%), and sodium hydroxide, NaOH (Prolabo, 97%), at 220 °C for 48 h. Copper nitrate (12.9 mmol, 3.05 g) and sodium molybdate (8.6 mmol, 2.08 g) were separately dissolved in 15 mL of water. The solutions were mixed to give an amorphous light-green precipitate. Sodium hydroxide (8.6 mmol, 0.34 g) dissolved in 10 mL of water was then added, yielding a color change to yellow-green. The suspension was poured into Teflon-lined stainless steel bombs (125 cm3), filled to ∼1/3, and heated at 220 °C for 2 days. The powder was filtered and then washed with water, followed by acetone, before being dried at 40 °C in air. XRPD indicates a pure sample where all the peaks were indexed to those of lindgrenite. Anal. Calcd for H2O10Cu3Mo2: Cu, 35.01; Mo, 35.24%. Found: Cu, 35.9; Mo, 35. 4%. A Sample of Cu3(OD)2(MoO4)2 (2) for powder neutron diffraction studies was prepared using the same synthetic procedure by replacing H2O with heavy water, D2O, to replace as much H as possible with D in the samples. Deuterium reduces the intense background usually observed for H samples which results from the high incoherent scattering factor of hydrogen. The powder was filtered; then, it was washed with a small amount of heavy water, followed by acetone, before being dried at 40 °C in air. Upon Rietveld refinement of the neutron diffraction data, we realized that the sample was contaminated with ∼3% Na2MoO4·2D2O, possibly because of inadequate washing with heavy water. Characterization. Thermal analyses were performed using a TASTD-Q600 apparatus at a 5 °C min-1 heating rate under air. Infrared spectra were recorded with a Digilab FTS3000 spectrometer by transmission through KBr pellets containing 1% of the powder. Powder X-ray diffraction patterns were recorded using a D5000 Siemens diffractometer equipped with a forward monochromator (Cu KR1, 1.5406 Å). The neutron diffraction experiments were performed at the Laboratoire Le´on Brillouin (CEA Saclay) using the 3T2 and G4.1 diffractometers. Data obtained on the high-resolution powder diffractometer 3T2 (λ ) 1.2252 Å, 6° < 2θ < 126°) were used for the refinement of the nuclear structure at 300 K, and those from the multidetector (800 cells) G4.1 (λ ) 2.4266 Å) were used for the determination of the magnetic structure and the thermal evolution study of the low-temperature patterns. Therefore, 10 diffraction patterns were recorded in the 2θ range of 11-90.9° at different temperatures between 1.5 and 18 K. The powder sample was placed in a cylindrical vanadium container and held in a liquid helium cryostat. Nuclear and magnetic structures were refined using the FULLPROF program.29 The nuclear scattering lengths (bCu ) 0.7718 × 10-12 cm, bMo ) 0.6715 × 10-12 cm, bO ) 0.5803 × 10-12 cm, bD ) 0.6671 × 10-12 cm, and bH ) -0.3739 × 10-12 cm) and copper magnetic form factors were included in this program. Magnetization measurements were performed in the range of 2-300 K and in a field of (50 kOe with a Quantum Design (28) Clearfield, A.; Moini, A.; Rudolf, P. R. Inorg. Chem. 1985, 24, 46064609. (29) Rodriguez-Carvajal, J. FULLPROF: RietVeld, profile matching and integrated intensity refinement of X-ray and/or neutron data, version 3.5d; Le´on-Brillouin Laboratory: CEA Saclay, France, 1998.

10940 Inorganic Chemistry, Vol. 45, No. 26, 2006

MPMS-XL SQUID magnetometer; ac susceptibility was determined in the same apparatus in an applied field of 3.5 Oe, oscillating at 20 Hz. Specific-heat measurements were carried out in a homebuilt apparatus employing a quasi-adiabatic method in the temperature range of 1.6-37 K.

Results and Discussion Synthesis. Although our original aim was to study the two minerals, lindgrenite and szenicsite,13 as a comparative work and with antlerite because of their slightly different structures, we have so far not succeeded in obtaining szenicsite. We tried to prepare szenicsite, Cu3(OH)4MoO4, using the same procedure described in the Experimental Section and by varying the experimental conditions. The conditions described in the experimental section principally produce lindgrenite with a yield of ∼90%, and XRD reveals the presence of a single phase. However, other phases have been identified under slightly different conditions. When the stoichiometric proportion Cu/Mo/NaOH is 3:2:2 pure lindgrenite results; the concomitant formation of NaCu(OH)MoO4 as a very minor phase is detected if the amount of NaOH is decreased to 1.5 and 1. In the presence of an excess MoO42-, Cu/Mo/NaOH ) 3:3:1-3, a gray or black precipitate is obtained, where XRD reveals a mixture of lingrenite and CuO or pure CuO. This result may be related to the increase of pH promoted by the excess of Na2MoO4·2H2O. In the presence of a low concentration of MoO42-, Cu/Mo/NaOH ) 3:0.5-1.5:0.5-1.5, pure lindgrenite is obtained for low concentrations of NaOH, and the concomitant formation of NaCu(OH)MoO4 is shown for higher concentrations. The starting copper salt, together with control of the alkalinity, is important. We tried several experiments in which we substituted copper sulfate for copper nitrate. If NaOH is also used, we always get CuO. Subsequently, experiments were performed without NaOH. Lindgrenite is obtained for Cu/ Mo ) 3:2.5-4.5, where the excess of Na2MoO4·2H2O is necessary to increase the pH value. For Mo contents higher than 4.5, NaCu(OH)MoO4 is obtained as dark-green single crystals. For all the above synthetic procedures, the temperature must be restricted to 220 °C; higher temperature always results in the formation of copper oxide. The slight differences for the two starting materials may be related to the respective pH values of copper nitrate and copper sulfate solutions, the first one being more acidic. Our procedure described above is closely related to those in the literature.27-28 Synthetic lindgrenite was previously obtained either by reflux overnight of a pale blue-green aqueous suspension obtained by dropwise addition of a Na2MoO4·2H2O solution to a copper sulfate(I)10,27,28 or by hydrothermal treatment of MoO3 and CuO in an aqueous Na2MoO4·2H2O solution at 200 °C30 or of an aqueous suspension obtained from (NH4)6Mo7O24·4H2O/CuCl2·2H2O/ 2-ampy at 180 °C for 2 days (2-ampy ) 2-aminopyridine).31 X-ray Powder Diffraction. X-ray diffraction data collected on the pale green powder indicate a highly crystalline (30) Moini, A.; Peascoe, R.; Rudolf, P. R.; Clearfield, A. Inorg. Chem. 1986, 25, 3782-3785. (31) Pavani, K.; Ramanan, A. Eur. J. Inorg. Chem. 2005, 3080-3087.

Magnetic Structure and Properties of Lindgrenite

materials, and indexing of the peaks confirmed the presence of only lindgrenite, Cu3(OH)2(MoO4)2 (JCPDS file 86-2311). All the Bragg reflections have been indexed using the pattern matching module of the FULLPROF program yielding the unit cell parameters a ) 5.6139(1) Å, b ) 14.0287(3) Å, c ) 5.3972(1) Å, and β ) 98.496(1)°, confirming that the sample is single phase.31 Figure S1 (Supporting Information) gives the corresponding experimental and calculated diffractograms. Thermal Analysis. The TGA (Figure S2) recorded under a flow of air reveals a weight loss between around 370 and 410 °C. It corresponds to the departure of OH groups as H2O, ∆Pobsd/P ) 3.27%, ∆Pcalcd/P ) 3.34%. The X-ray diffraction pattern recorded on the sample, having a red color, obtained after this weight loss reveals the formation of Cu3Mo2O9 in agreement with JCPDS file 24-55. This agreement between calculated and observed weight loss and the content of the residual compound, Cu3Mo2O9, having the same stoichiometry as lindgrenite, confirms the purity of 1. A second weight loss takes place between around 840 and 895 °C and is related to the decomposition of Cu3Mo2O9 into unidentified oxides. Infrared Spectroscopy. The IR spectrum recorded between 300 and 4000 cm-1 on 1 reveals three families of vibrational bands centered around 3350, 850, and 440 cm-1. As the first one is shifted in the spectrum of 2, it can be attributed to the symmetric and antisymmetric stretching modes of the O-H bonds. In 2, these bands appear at around 2500 cm-1 [i.e., a ratio of 0.75 which is close to the expected value of x(18/(2 × 17))]. Between 960 and 820 cm-1, four bands (one broad, two sharp. and one shoulder) are observed for 1 and 2. The ν1 and ν3 modes of the molybdate ion are expected in this region, where ν1 ) 897 cm-1 and ν3 ) 837 cm-1 were observed for the free ion in solution.32 Because of the C1 local symmetry, the number of infrared active vibrations increases to 3 for ν3, and ν1 becomes infrared active. Because the relative intensities of the bands between 960 and 820 cm-1 do not significantly vary between 1 and 2, they have been attributed to MoO42- vibrations. We propose the assignment of the 958 cm-1 band to ν1(MoO4) and the 914, 868, and 825 cm-1 bands to ν3(MoO4) in agreement with the proposed assignment of ν1 at the highest wavenumber.32 These bands appear at 952, 911, 865, and 831 cm-1 for the deuterated sample, while only the last three are observed for Na2MoO4·2H2O possibly because of the higher symmetry of the MoO4. The spectrum of 2 reveals two additional sharp bands at 671 and 635 cm-1 that have been attributed to the libration modes. Therefore, the corresponding O-H bands are expected in the same regions as those of MoO42-; the O-H libration bands probably being unresolved, are hidden by the broad band at 825 cm-1.32 The five bands between 460 and 300 cm-1 have been attributed to (a) Cu-OH(D) vibration bands at 455 and 440 cm-1 for (32) (a) Frost, R. L.; Duong, L.; Weier, M. Neues. Jahrb. Mineral. 2004, 180, 245-260. (b) Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coordination Compounds; John Wiley: New York, 1986. (c) Weinstock, N.; Schulze, H.; Mu¨ller, A. J. Chem. Phys. 1973, 59, 5063-5067.

Table 1. Summary of the Neutron Powder Diffraction Data Analyses for Nuclear and Magnetic Structures

a (Å) b (Å) c (Å) β (deg) V (Å3) Z space group wavelength (Å) 2θ range, step reflns (nuclear) reflns (magnetic) params RP (%) RWP (%) RB (%) RF (%) Rmagnetic (%) GOF

300 K

1.5 K

5.3980(2) 14.0234(4) 5.6129(2) 98.497(2) 420.22(5) 2 P21/n 1.2244 6.08-120.63°, 0.05° 1310

5.4052(2) 14.0716(5) 5.6183(2) 98.745(2) 422.36(6) 2 P21/n 2.4266 11.0-90.9°, 0.1° 99 197 39 7.27 8.41 2.00 1.64 4.99

45 5.34 6.26 1.49 0.70 2.04

Table 2. Final Atomic Positions from Neutron Powder Diffraction Data at 300 K atoma Cu1 Cu2 Mo O1 O2 O3 O4 OD H D a

x/a

y/b

z/c

B (Å)

occupation

0 0.8659(4) 0.4559(4) 0.9805(6) 0.6541(6) 0.5543(6) 0.1447(6) 0.1378(6) 0.2874(6) 0.2874(6)

0 0.0947(2) 0.1547(2) 0.2226(2) 0.0928(2) 0.1307(2) 0.1143(2) 0.0311(2) 0.0684(2) 0.0684(2)

0 0.4881(4) 0.8763(4) 0.4412(5) 0.1000(5) 0.5948(5) 0.8767(5) 0.3448(5) 0.3724(6) 0.3724(6)

0.88(5) 0.96(4) 0.54(3) 1.11(5) 1.29(5) 1.28(5) 1.04(5) 1.08(5) 1.46(6) 1.46(6)

2.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 0.36(3) 3.64(3)

D/(H + D) ) 0.91.

1 and 2, respectively, the slight shift toward lower energies for 2 being in agreement with the mass difference (such a band is not shown in the IR spectrum recorded under the same conditions on Na2MoO4·2H2O), (b) to O-H vibration at 420 cm-1 since this band nearly disappears for 2, and (c) to ν2 and ν4(MoO4) for the bands at 387, 352, and 341 cm-1, their positions being the same for 1 and 2. The results are summarized in Table S1 and Figure S3 (Supporting Information). Nuclear Structure Description. Although the nuclear structure of the natural mineral is known, it is important for this work to show that the synthetic analogue has the same structure at all temperatures, as well as establish the hydrogen positions precisely from the neutron data. Therefore, the nuclear structure has been refined (Tables 1 and 2) from the neutron powder data collected at room temperature on the high-resolution 3T2 diffractometer and from the X-ray powder diffraction starting from the atomic coordinates reported by Hawthorne and Eby (Figure S4).14 A good agreement with the X-ray single-crystal results is observed in both cases. Three very weak lines of the G4.1 neutron data were not indexed and have been attributed to the presence of a small amount of sodium molybdate, Na2MoO4· 2D2O. It was not observed by X-ray diffraction, and it amounts to around 3% according to the refinement. The inclusion of its contribution in the refinement through the pattern-matching procedure results in a better convergence. As the main features of the structure are already reported,14 Inorganic Chemistry, Vol. 45, No. 26, 2006

10941

Vilminot et al.

Figure 1. (a) View of a single ribbon showing the edge-sharing copper octahedra where the long bonds characterizing the Jahn-Teller distortions are highlighted by yellow oxygen atoms and (b) a projection of the structure along the ribbons (c axis) showing the herringbone arrangement and the connection of the ribbons by the MoO4.

Figure 2. The µ6-mode of coordination of the MoO4 unit to its three neighboring ribbons of copper hydroxide.

Table 3. Selected Interatomic Distances (Å) at 300 K Cu1-O2 Cu1-O4 Cu1-OD Mo-O1 Mo-O2 Mo-O3 Mo-O4

2.409(3) (×2) 1.953(3) (×2) 2.016(3) (×2) 2.126 1.760(4) 1.754(4) 1.772(4) 1.774(4) 1.765

Cu2-O1 Cu2-O2 Cu2-O3 Cu2-O4 Cu2-OD Cu2-OD OD-D D‚‚‚O3

1.928(4) 2.306(3) 1.934(4) 2.476(3) 1.986(4) 1.999(4) 2.105 0.956(4) 1.966(4)

we will only highlight some of the features relevant to the magnetism and magnetic structure discussed below. The key feature of the structure of lindgrenite is the presence of Cu3(OH)2 ribbons (Figure 1a) of the brucite structure consisting of edge-sharing copper octahedra running parallel to the c axis (Figure 1b). There are two kinds of copper atoms, Cu(1) and Cu(2), both having distorted octahedral coordination of oxygen atoms. The MoO4 tetrahedra connect these chains into a 3D net where the ribbons form a herringbone arrangement when viewed down the ribbon axis. Each ribbon alternately contains one Cu(1) octahedron and two Cu(2) octahedra, all being edge-shared (Figure 1). The copper octahedra are strongly distorted with a 4 + 2 environment (Tables 3 and S2). For both copper atoms, their environments contain four oxygen atoms from molybdate anions and two hydroxides. While for Cu(1), the OD groups are in trans positions, they are in cis positions for Cu(2). The apical distortion of the copper octahedra is related to a strong Jahn-Teller effect (Figure 1a). The distortion is more symmetric for Cu(1) with two equal apical bonds, 2.409 Å, and a mean equatorial bond length of 1.985 Å because Cu(1) lies on a center of symmetry. For Cu(2), the corresponding bond lengths are 2.306 and 2.476 Å for the apical ones and 1.962 Å for the mean equatorial value. These distortions are also seen in the angles that range from 73 to 99° and from 155 to 180° for O-Cu-O. The MoO4 plays an important role in the structure, as does the SO4 in the metal-hydroxy sulfates.2 It connects three ribbons together (Figure 2) via its oxygen atoms: O(1) and O(3) are bonded to single Cu atoms, while O(2) and O(4) bridges two copper atoms each on two adjacent chains. The latter oxygen atoms are also part of the skeleton of the ribbons. The MoO4 shares its oxygen corners with six copper atoms defining a µ6 environment. The MoO4 tetrahedron is

10942 Inorganic Chemistry, Vol. 45, No. 26, 2006

Table 4. Empirical Bond Valence for Lindgrenite33 Cu1

Cu2

Mo

O1 O2 O3

0.139(1) (×2)

0.510(5) 0.184(2) 0.502(5)

1.489(14) 1.511(14) 1.441(15)

O4 OD

0.477(4) (×2) 0.403(3) (×2)

1.434(15)



2.036(7)

0.116(1) 0.436(5) 0.422(4) 2.169(10)

5.875(29)

H+D

0.020(1) 0.055(1) 0.060(3) 0.842(3) 0.977(3)

∑ 1.999(15) 1.834(15) 2.018(16) 2.026(15) 2.163(14)

slightly distorted (Tables 3 and S2) with Mo-O bond lengths between 1.754 and 1.774 Å (mean value 1.765 Å) and angles between 108.1 and 111.7°. Its deformation from a regular tetrahedron is less pronounced compared to the X-ray results of Hawthorne and Eby.14 The O-D bond length is 0.956(4) Å, and the group is involved in one deuterium bond with O3, D-O3 ) 1.966(4) Å and OD-D-O3 ) 150.3(5)°. The oxygen atoms exhibit µ2 (O1 and O3) and µ3 (O2, O4, OD) environments. Finally, calculation of the bond valences (Table 4) according to the model of Altermatt and Brown yields values in agreement with the expected ones.33 The agreement is even better than for the corresponding values calculated from the X-ray single-crystal data of Hawthorne and Eby.14 This is particularly true for the O4 atom where the sum of individual bond valences decreases from 2.53 to 2.026. Magnetic Properties. The magnetic properties of the two compounds Cu3(OH)2(MoO4)2 (1) and Cu3(OD)2(MoO4)2 (2) are very similar, and here, we report those of the purer of the two, Cu3(OH)2(MoO4)2 (1). The magnetic susceptibility measured in an applied field of 100 Oe is shown in Figure 3 as 1/χ and χT. From 300 to 30 K, χT decreases gradually (33) (a) Altermatt, D.; Brown, I. D. Acta Crystallogr. 1985, B41, 240244. (b) Brown, I. D.; Altermatt, D. Acta Crystallogr. 1985, B41, 244247. (34) Kurmoo, M.; Kumagai, H.; Akita-Tanaka, M.; Inoue, K.; Takagi, S. Inorg. Chem. 2006, 45, 1627 and references therein. (35) Bertaut, F. Acta Crystallogr. 1968, A24, 217. (36) (a) Hovestreydt, E.; Aroyo, I.; Sattler, S.; Wondratschek, H. KAREPs A program for calculating irreductible space group representations. J. Appl. Crystallogr. 1992, 25, 544. (b) Rodriguez-Carvajal, J. BASIREPSs A program for calculating non-normalized basis functions of the irreductible representations of the little group Gk for atom properties in a crystal; Laboratoire Le´on Brillouin (CEA-CNRS): CEA Saclay, Gif sur Yvette, France.

Magnetic Structure and Properties of Lindgrenite

Figure 3. Temperature dependence of the inverse magnetic susceptibility (red circles) and of effective moment represented as χT (blue circles). The black solid line is the least-square best fit to the Curie-Weiss law.

Figure 4. Temperature dependence of the dc and ac magnetic susceptibilities.

to a shallow minimum before a sharp increase, reaching a maximum at 10 K. In the temperature range of 100-300 K, the magnetic susceptibility follows the Curie-Weiss law, χ ) C/(T - θ), with the Curie constant of C ) 1.406(1) emu mol-1 and the Weiss temperature of θ ) -14.2(2) K. Using C ) Ng2µB2S(S + 1)/3k and S ) 1/2 for Cu2+, we obtain g ) 2.236(1), which is in fair agreement with the values in the literature. The negative Weiss temperature suggests the dominance of antiferromagnetic near-neighbor interaction. Below 28 K, the χT product starts to increase, first progressively and then steeply between 13 and 12 K to reach a maximum at 10 K. At lower temperatures, χ reaches a plateau, whereas χT decreases. The ac-susceptibility measurements reveal the presence of a peak on both the real and imaginary components with a maximum of χ′ at 12.6 K (Figure 4), signaling the presence of a long-range magnetic ordering. It is worth noting that the degree of frustration (f),37c estimated by the ratio of the Weiss temperature (θ) to the Curie or Ne´el temperature (TC or TN), is almost unity. This may then be regarded as being close to the mean field, but in the presence of both ferromagnetic and antiferromagnetic exchanges, it cannot be the case (see later). A second anomaly is also detected at around 8 K. Isothermal magnetization at different temperatures (Figure 5) were recorded (37) (a) Riuz, E; Cano, J.; Alvarez, S.; Alemany, P. J. Am. Chem. Soc. 1998, 120, 11122. (b) Weihe, H.; Gu¨del. H. U. J. Am. Chem. Soc. 1997, 119, 6539. (c) Weihe, H.; Gu¨del, H. U. J. Am. Chem. Soc. 1998, 120, 2870. (d) Weihe, H.; Gu¨del, H. U. Inorg. Chem. 1997, 36, 3632.

Figure 5. Isothermal magnetization at different temperatures.

to shed some light on the latter anomaly. First, we observe a paramagnetic behavior above (18 K) the Curie temperature and superparamagnetic just above (13.4 K). At 12.6 K, there is a small hysteresis loop with a very weak coercive field. As the temperature is lowered further, a prominent square shape hysteresis loop develops whose coercive field and remanant magnetization increase (Figures S5 and S6). The magnetization reaches a value of 1.256 N µB mol-1 (0.42 N µB mol-1 of Cu) at 2 K in the maximum field. Therefore, all the results (minimum in χT, negative Weiss temperature, and saturation magnetization of nearly 1/3 of the expected total) are consistent with those expected for a topological ferrimagnetic chain differing in the number of equivalent moments in the two sublattices.34 In addition, the sharp anomalies at 12.6 K in the ac susceptibilities confirm the long-range ordering to be ferrimagnetic. One unusual observation is the constant increase of the saturation magnetization without a change in slope of the magnetization as a function of field (>20 kOe) for all temperatures below TC. The value of the saturation magnetization, of nearly 1/3 of the total moment, is close to that expected for one resultant copper moment. However, the hysteresis loop measurements in the temperature range, where the second anomaly at 8 K was observed in the ac-susceptibility data, do not reveal any change in the evolution. It is to be noted that there is a decrease in the real part of the ac susceptibilities, suggesting that the anomaly cannot be the presence of an impurity. A magnetic impurity is expected to add a component to the total. One plausible explanation is a change of magnetic hardness or relaxation intrinsic to the low-dimensional nature of the compound. Consequently, the anomaly is not seen in the dc magnetization. With the exception of the ac susceptibilities and the isothermal magnetization at different temperatures that were not reported by Shores et al., all our results are in good agreement.10 Specific Heat. To get more insight of the magnetic ordering for this material, the specific heat measurements in zero-field were performed. Figure 6 shows the experimental temperature dependence of the temperature-normalized heat capacity corrected for the lattice phonons as follows: Cm ) Cp - Clattice, where the lattice contribution is expressed by Clattice ) β1T3 + higher terms (higher terms are assumed to be very small in our case because of the Inorganic Chemistry, Vol. 45, No. 26, 2006

10943

Vilminot et al. Table 5. Basis Functions of the Irreducible Representations, Γ2 and Γ4a IR

moment

Cu11

Cu12

Cu21

Cu22

Cu23

Cu24

Γ2 Γ4

Mx, My, Mz Mx, My, Mz

+++ +++

-++-+

+++ +++

+++ +++

-++-+

-++-+

a Cu11 (0, 0, 0); Cu12 (1/2, 1/2, 1/2); Cu21 (x, y, z); Cu22 (-x, -y, -z); Cu23 (1/2 - x, 1/2 + y, 1/2 - z); Cu24 (1/2 + x, 1/2 - y, 1/2 + z).

Figure 6. Temperature dependence of magnetic heat capacity normalized to the temperature (experimental data in red circles) and the corresponding integrated entropies (blue circles). Lines are a guide for the eyes.

limited temperature range of the measurements) and β1 is related to the Debye temperature. The magnetic heat capacity normalized by the temperature shows a λ anomaly, peaking at 12.6 K, at the long-range magnetic ordering discussed above. The magnetic entropy was obtained by integrating the Cm/T versus T curve within the limit of the experimental temperature range (assuming the heat capacity is zero at 0 K). The value of the total entropy up to 37 K of 17.7 J K-1 mol-1 is close to that expected for three copper atoms, 3R ln(2S + 1), of 17.3 J K-1 mol-1 for S ) 1/2. Of this total, only 50% is accounted for below the Curie temperature. The remaining entropy is, therefore, associated with the short-range correlation resulting from the low-dimensionality of the system. Again no unusual behavior is observed at 8 K. The fact that (a) there is a welldefined λ-peak in the heat capacity data at the long-range magnetic ordering temperature, (b) the heat capacity tends toward zero as the temperature moves toward zero, and (c) the estimated magnetic entropy is as expected rules out the presence any geometry-associated magnetic frustration. This agrees with the frustration factor of unity from the magnetic data discussed above. Magnetic Structure. A comparison of the diffractograms (λ ) 2.4266 Å) recorded at 1.5 and 17.8 K reveals only slight differences as expected for the low moment of copper. No new lines appear, and one observes principally an enhancement in the intensity of the line at 2θ ) 27.17°, indexed as (0 1 1) (Figures S7 and S8). The same observation is made for the (-1 0 1) line at 33.73°. These Bragg angles of the difference peaks are consistent for all temperatures below the long-range magnetic ordering suggesting a commensurate k ) (0 0 0) propagation vector. In the difference pattern, the intensities of few other lines increase very slightly at low temperature. Among them, one notices the (0 4 0) at 40.35° and a weaker one (0 0 2) at 51.83°. Such behavior is usually characteristic of a ferromagnetic structure since all the magnetic scatterings appear at the same angles as those of the nuclear ones. The fact that the (0 4 0) and (0 0 2) lines have each a magnetic contribution suggests the most possible orientation of the magnetic moments is along the a axis. To solve the magnetic structure, we determined the associated magnetic structure with the help of Bertaut’s representation analysis method,35 applied to P21/n space

10944 Inorganic Chemistry, Vol. 45, No. 26, 2006

Figure 7. Observed (circles) and calculated (line) profiles of the neutron powder diffrraction pattern of Cu3(OD)2(MoO4)2 obtained on the G4.1 diffractometer (λ ) 2.4266 Å) at 1.5 K with position of the Bragg reflections (short vertical lines) of Cu3(OD)2(MoO4)2 and of Na2MoO4‚D2O and difference between observed and calculated profiles. Table 6. Results of the Refinements moment orientation a axis a axis b axis b axis

copper ion Cu1 Cu2 Cu1 Cu2 Cu1 Cu2 Cu1 Cu2

Mx (µB) 0.83(30) -0.78(19) 0.77(4) -0.77(4) 0a 0a 0a 0a

My (µB) 0a 0a 0a 0a 0.81(26) -0.77(16) 0.79(4) -0.79(4)

Rmagnetic 5.45% 4.90% 6.38% 6.33%

a The moments are significantly lower than their corresponding standard deviations. The same is true for the Mz component for all models.

group, k ) (0 0 0) propagation vector, and 2a and 4e Wyckoff positions for the Cu1 and Cu2 copper ions, respectively. Four one-dimensional irreducible representations (IR), Γ1-Γ4, are associated to basis vectors (magnetic structures) for Cu2, while only two, Γ2 and Γ4 (Table 5), are associated with Cu1.36 Therefore, only the latter two need to be considered as possible models. In the first one, the moments are along the b axis; in the second one, they are in the ac plane. Both representations have been checked by considering that the Cu1 and Cu2 magnetic moments are either independent or equal. As expected, the best refinement (Figure 7) is obtained for the Γ4 representation with the magnetic moments oriented along the a axis, with opposite orientation of moments for Cu(1) and Cu(2). The refinement performed for independent moments of the two copper atoms was of lower certainty compared to those for equal moments. On the other hand, a model with moments along the b axis is less reliable (Table 6). From this table, it appears that the best model corresponds to magnetic moments along the a axis with equal values for

Magnetic Structure and Properties of Lindgrenite

Figure 8. Alignment of the magnetic moments in the magnetic structure of Cu3(OH)2(MoO4)2 below 12 K.

Figure 9. Temperature dependence of the magnetic moment per copper.

both Cu1 and Cu2. The presence of a magnetic contribution on the (0 4 0) line certainly confirms this model. Figure 7 shows the observed and calculated diffraction patterns. The magnetic structure of lindgrenite is that of a topological ferrimagnet (Figure 8) with magnetic moments aligned along the a axis for Cu(1) and in the opposite one for Cu(2) and the moment value being 0.77(4) µB at 1.5 K (Figure 9). The same model also applies for all the temperatures below TC, and the thermal evolution of the magnetic moments is given in Figure 9. The magnetic moment value is nearly constant between 1.5 and 8 K and starts to decrease at 10 K (Table S3). We note that the moment saturates around the temperature where the anomaly is observed in the ac susceptibilities. No other difference is observed in the diffractograms. The fact that the model with equal moments gives rise to a better fit is because the two copper atoms have similar coordination geometry. It is also worth noting the absence of any broad diffraction and enhanced scattering at small angles at all temperatures, above and below the long-range magnetic ordering; this may indicate the presence of frustration or short-range clustering of moments.2 It is of interest to understand why lindgrenite orders as a ferrimagnet by searching for any existing structural correlation. For this 3D structure, there are two crystallographically independent copper atoms, namely, Cu(1) and Cu(2), where there are one Cu(1) and two Cu(2) in a unit belonging to a ribbon. Cu(1) sits on an inversion center. The unit cell

contains two of Cu(1) and four of Cu(2) belonging to a central ribbon and four near neighbor ones. The closest CuCu distance (3.01-3.21 Å, Table S2) is via one-oxygenatom bridges where the oxygen is either the OH, O(2), or O(4) of MoO4. These provide the major magnetic interaction pathways. Because of the three different bond-distances and the Cu-O-Cu angles, one can define three independent interactions. Consequently, the expected degeneracy arising from geometrical frustration of an equilateral triangular system of S ) 1/2 can be lifted. Assuming the results of existing theoretical calculations that for an angle less than 98° a ferromagnetic exchange is favorable and one greater than 98° is antiferromagnetic, we can define both for the nearest-neighbor copper atoms. We note that there are two different angles defining each magnetic pathway (Table S2), and in most cases, one angle is lower than 98° and the other is greater. It is believed that in such cases the AF is the most likely exchange.37 The next hierarchical exchange interactions are those involving Cu-O-Mo-O-Cu pathways. Here, we can define 13 independent ones, 2 within one ribbon and 11 between ribbons. The MoO4 units around one ribbon connect six neighboring ribbons. In principle, these are very weak and appear to favor parallel alignment of the corresponding moments on neighboring ribbons. Therefore, they must be weakly ferromagnetic in nature. It is interesting to note that the weak magnetic anisotropy of copper favors a moment perpendicular to the ribbon, while a strong anisotropy (from say Co(II) or Mn(III)) may favor an Ising system with alignment along the ribbons. The above arguments and the results of magnetization and neutron scattering are consistent with a picture where the moments of the two different copper atoms are aligned antiparallel to one another within the ribbons with the resultant moment of each ribbon being parallel to give a ferrimagnet. This is different than the interpretation and proposed mechanism of the magnetization reversal by Shores et al.10 From the results of our findings, it is worth noting in the work of Shores et al. that introduction of bipyridine or piperazine separates the 3D structure into layers and, consequently, results in an antiferromagnetic ground state, where the weak antiferromagnetic interaction via the ditopic ligands can be switched by a weak magnetic field of 2 kOe to obtain the ferrimagnetic ground state of the parent lindgrenite. Although the fitting of high-temperature magnetic susceptibility data for compounds having bpy bridges suggest that exchange through the bpy ligand is zero,38 the data of Shores et al.10 and those of more recent studies39 show that exchange through bpy is present but may be small. In contrast, a different picture emerges for the presence of the more anisotropic Co2+ in Co3(OH)2(C4O4)2· 3H2O,15 where each ribbon is ferromagnetic and the moments on adjacent ribbons are antiparallel, thus resulting in a LRO Ne´el ground state in a small field. The application of a high (38) Julve, M.; Verdaguer, M.; Faus, J.; Tinti, F.; Moratal, J.; Monge, A.; Gutie´rrez-Puebla, E. Inorg. Chem. 1987, 26, 3520. (39) (a) Wang, X-Y.; Wei, H-Y.; Wang, Z-M.; Chen, Z-D.; Gao, S. Inorg. Chem. 2005, 44, 572-583. (b) Feyerherm, R.; Loose, A.; Lawandy, M. A.; Li, J. Appl. Phys. A 2002, 74 (Suppl., Part 1), S778-S780. (c) Feyerherm, R.; Loose, A.; Lawandy, M. A.; Li, J. Phys. Chem. Solids 2002, 63, 71-77.

Inorganic Chemistry, Vol. 45, No. 26, 2006

10945

Vilminot et al. field reverses the moment easily to the ferromagnetic state. Furthermore, removal of the occluded water molecules in the channels also results in a ferromagnetic state. This process is reversible. The results of these works highlight the importance of weak magnetic exchange through large organic bridges or hydrogen bonds that can transform these materials between different ground states. Conclusion Cu3(OH)2(MoO4)2 and Cu3(OD)2(MoO4)2, prepared at moderately low temperatures in the laboratory, are isostructural to the mineral lindgrenite. The magnetic properties are consistent with those of a ferrimagnet with a Curie temperature (TC) of 13 K with reasonable magnetic hardness with complete absence of geometrical frustration. The heat capacity gives only 50% of the entropy below TC, and the remaining amount is taken up by the low dimensionality. Below TC, the magnetic structure refined from neutron diffraction data presents ferrimagnetic ribbons with parallel resultant moments. The distortion of the Cu3(OH) unit from

10946 Inorganic Chemistry, Vol. 45, No. 26, 2006

being an equilateral triangle appears to lift the degeneracy in the spectrum, thus resulting in an ordered state without spin frustration.40 Using the present result from the neutron data analyses of the parent lindgrenite, we were able to propose a different model to explain the magnetic properties of the bipyridine-functionalized lindgrenite.10 Acknowledgment. We gratefully acknowledge the CNRS for funding. Supporting Information Available: Infrared, DT-TGA, nuclear and magnetic crystallographic (tables of bond lengths and angles), and magnetic data for compounds 1 and 2. This material is available free of charge via the Internet at http://pubs.acs.org. IC061182M (40) (a) Bramwell, S. T.; Gingras, M. J. P. Science 2001, 94, 1495. (b) Greedan, J. E. J. Mater. Chem. 2001, 11, 37. (c) Ramirez, A. P. In Handbook of Magnetic Materials; Buschow, K. J. H., Ed.; Elsevier Science: Amsterdam, 2001; Vol. 13, p 423. (d) Grohol, D.; Matan, K.; Cho, J.-H.; Lee, S.-H.; Lynn, J. W.; Nocera, D. G.; Lee, Y. S. Nat. Mater. 2005, 4, 323. (e) Diep, H. T., Ed. Magnetic Systems with Competing Interactions: Frustrated Spin Systems; World Scientific: Singapore, 1994.